A classroom day necessarily aims at the middle. Exactly the two groups you wanted to reach fall outside it.
In every team we measure there's someone who has worked with AI daily for months and someone who has never opened it. Usually three more levels sit in between. A trainer facing that group has one choice to make and it's always wrong: aim high and the bottom drops out; aim low and the top gets bored.
That isn't a criticism of trainers. It's arithmetic. A room has one pace and a group has five.
What it costs the frontrunners
The group that's already ahead is the one you need most. They pull colleagues along, answer questions by the coffee machine and provide the example that convinces the sceptics. Have them watch an introduction for a morning and you lose exactly that, and it's nearly impossible to get back.
What does work
- Modules at your own pace, so whoever knows it clicks through and whoever doesn't rewatches.
- A measurement first, so you know who sits where instead of guessing.
- A separate session for the frontrunners, where they build something instead of listening.
- Classroom only for what needs a room: resistance, agreements, deciding together.
That last line is the most important, and it argues against our own platform. There are topics for which a screen simply isn't enough. There are just fewer of them than the standard approach assumes, and you can only identify them after measuring.